ABSTRACT
Conventional views understand assemblage variation in essentialist terms that correspond to ethnographic patterns at various levels. In North American Palaeoindian data, however, some variation is categorical but much is continuous and some is determined by assemblage size, a factor more often controlled than studied. A similar view emerges in the study of 146 Middle Palaeolithic assemblages from Europe and the Near East. There, degree of tool reduction also explains some variation in proportional frequency of tool 'types'. Despite obvious differences of time and place, similar formation processes operated in the cases. Norm and activity contribute to assemblage variation, but only among other factors like formation processes.
